| 摘要: | 本論文以生成式人工智慧在創作實務中的應用為出發點，探討現行著作權制度在面對AI協助創作時所產生之法律定位與制度挑戰。隨著生成式AI已廣泛被創作者作為創作工具使用，現行以人類著作人原則為核心的著作權制度，逐漸面臨是否承認、以及如何承認AI協作創作成果之保護問題。
本文首先整理著作權法中關於作者、原創性與創作主體之基本理論，並回顧人工智慧出現前，司法實務與學說對非人類創作之處理共識，作為後續分析之基礎。其次，透過美國近年涉及人工智慧創作之重要案例與著作權局實務，包括A Recent Entrance to Paradise、Zarya of the Dawn、Théâtre D’opéra Spatial與SURYAST等案，分析現行實務對AI生成內容所採取之審查標準與其侷限。
本文主張，生成式人工智慧應定位為創作輔助工具，而非獨立創作主體。在使用者對創作過程與最終成果具有實質創作控制與原創性參與之前提下，人工智慧協作創作之作品不應僅因技術介入即被一概排除於著作權保護之外。為回應實務判斷困難，本文進一步提出以創作參與與控制為核心之判斷架構，作為認定使用者是否具備著作人地位之參考標準。
最後，本文比較各國制度對人工智慧協作創作之回應，並反思我國現行著作權制度之適用可能性，認為在不另立新權利制度之前提下，透過行政指引與實務判準之補充，仍可回應生成式人工智慧創作所帶來之挑戰，兼顧創作自由、制度安定性與著作權法之立法目的。
;This thesis examines the legal status and regulatory challenges of copyright protection for works created with the assistance of generative artificial intelligence. As generative AI has increasingly been adopted by creators as a creative tool, copyright systems grounded in the principle of human authorship are confronted with growing uncertainty regarding whether, and under what conditions, AI-assisted works should be protected.
The thesis first reviews foundational copyright doctrines concerning authorship, originality, and creative subjectivity, and surveys pre-AI judicial and scholarly approaches to non-human creation as a baseline for analysis. It then analyzes recent U.S. copyright practice and administrative decisions involving AI-generated content, including A Recent Entrance to Paradise, Zarya of the Dawn, Théâtre D’opéra Spatial, and SURYAST, to identify prevailing evaluative standards and their practical limitations in addressing human–AI collaborative creation.
This thesis argues that generative AI should be understood as a creative assistive tool rather than an independent author. Where a human user exercises substantive creative control and makes original contributions throughout the creative process and to the final output, the resulting work should not be categorically excluded from copyright protection solely due to the involvement of AI technology. To address the difficulties faced in current practice, this thesis proposes a participation- and control-based analytical framework for assessing human authorship in AI-assisted works.
Finally, through a comparative review of international approaches to AI-assisted creation, this thesis evaluates the implications for Taiwan’s copyright system. It concludes that, without establishing a new sui generis right, existing copyright law can accommodate generative AI-assisted creation through the development of administrative guidelines and refined evaluative standards, thereby preserving creative freedom, legal certainty, and the fundamental objectives of copyright law. |
